Get the newsletter
Adobe · AI / ML · Filed Apr 8, 2026 · USPTO published Aug 20, 2026 · Aug 20 release

Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You

Adobe wants to take the guesswork out of email marketing by having an AI suggest which template to use, which headline variants to test, and which combinations are actually worth running, all before you hit send.

Graphical user interface of an email design application featuring template customization options. Drawing from patent filing US 2026/0244924 A1.
Graphical user interface of an email design application featuring template customization options.All 30 drawings ↓
Publication number
US 2026/0244924 A1(US20260244924A1)
Applicant
Adobe Inc.
Filed · Published
Apr 8, 2026 · Aug 20, 2026
Inventors
Eunyee Koh, Tak Yeon Lee, Andrew Thomson, Vasanthi Holtcamp, Ryan Rossi, Fan Du, Caroline Kim, Tong Yu, Shunan Guo, Nedim Lipka, Shriram Venkatesh Shet Revankar, Nikhil Belsare
Status when we published
Waiting for an examiner
Document
20 claims
US classification
705/14.4
Parent application
Division of 17/383,114 (filed Jul 22, 2021)(17383114)

What Adobe's AI-assisted email builder actually does

Ever stared at a blank email editor wondering whether your subject line will land? Most marketers end up picking a layout by gut feel, running a few test versions, and hoping for the best. Adobe's new patent describes a system designed to replace that guesswork with data.

Here is how it works in plain terms. You open the editor and the tool shows you a list of email templates, each labeled with a predicted performance score based on your past campaigns and the profiles of the people you are mailing. Pick one, and as you build out each section, headline, image, call-to-action button, the system suggests multiple versions of that section, again ranked by predicted performance. You pick your favorites, and the system then tells you which combinations are the most useful to test against each other.

The goal is to shrink the number of test versions you need to run. Instead of throwing twenty variants at your audience and waiting weeks for results, the AI surfaces the four or five combinations that are most likely to tell you something meaningful.

From the filing · CLAIM 1
… in response to a user interaction with a content fragment, providing, for display, a plurality of fragment variants for the content fragment according to predicted content fragment performance metrics for the plurality of fragment variants; …

Translation: Clicking a piece of content brings up alternative options ranked by predicted success.

How the system ranks templates, fragments, and test groups

The patent describes a multi-stage editorial interface for building and testing digital communications, with machine learning recommendations baked into each stage.

Stage one: template selection. The system displays a set of pre-built email layouts alongside predicted performance metrics (essentially a score estimating click rates, open rates, or another goal you care about). Those predictions draw on historical campaign data and recipient profile data (demographic and behavioral information about the people on your list).

Stage two: fragment-level editing. Once you pick a template, the email is broken into discrete content fragments, think of these as individually swappable blocks like a subject line, a hero image, or a footer. When you click any block, the system surfaces multiple fragment variants (alternative versions of that block), each labeled with its own predicted performance score.

Stage three: multivariate test planning. After you have selected your preferred variants across all the blocks, the system generates multivariate testing recommendations. Multivariate testing (running several variables at once to see which combination wins) usually produces an explosion of possible combinations. The patent's claim is that by pre-filtering variants using the model's predictions, the system narrows that list to only the statistically meaningful candidates, giving you cleaner results faster.

From the filing · THE ABSTRACT
Ultimately, the disclosed systems generate multivariate testing recommendations incorporating selected fragment variants to intelligently narrow multivariate testing candidates and generate more meaningful and statistically significant multivariate testing results.

Translation: The system narrows down testing options so you get more meaningful results.

What this means for marketers running email campaigns

For anyone managing email campaigns at scale, the practical win here is time. Multivariate testing is theoretically powerful but practically painful: if you have three content blocks and four variants of each, you are looking at 64 possible combinations. Most teams never test that many, so they either skip testing or run underpowered experiments that produce ambiguous results. A system that prunes that list to the most promising combinations before the campaign launches could make rigorous testing accessible to smaller teams who do not have a data scientist on staff.

The deeper strategic angle is that Adobe is positioning its creative tools not just as a canvas but as an optimization layer. Embedding performance predictions into the editing UI ties the act of creating content directly to the act of measuring it, which is exactly the workflow its marketing cloud customers need. Adobe's ad-tech and content-optimization work sits alongside a broader wave of AI-assisted campaign tools covered among the newest Big Tech patents, and this filing shows the company pushing that work deeper into the design phase itself.

Adobe's 119th filing in everything we've tracked since May joins earlier applications on fixing AI database code and building images from multiple photos.

Source: US 2026/0244924 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.

Get our take in your Top Stories

Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Adobe patent news breaks.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings · 30 sheets

30 drawing sheets from US 2026/0244924 A1 · click any drawing to enlarge

Networked system architecture overview, drawing sheet 1 from US 2026/0244924 A1
Sheet 1 Networked system architecture overview
Overview of multivariate testing workflow, drawing sheet 2 from US 2026/0244924 A1
Sheet 2 Overview of multivariate testing workflow
Multivariate testing system architecture, drawing sheet 3 from US 2026/0244924 A1
Sheet 3 Multivariate testing system architecture
Clustering historical communications by topic, drawing sheet 4 from US 2026/0244924 A1
Sheet 4 Clustering historical communications by topic
Fragment machine learning model architecture, drawing sheet 5 from US 2026/0244924 A1
Sheet 5 Fragment machine learning model architecture
Image content model descriptor determination, drawing sheet 6 from US 2026/0244924 A1
Sheet 6 Image content model descriptor determination
Multivariate testing results prediction model, drawing sheet 7 from US 2026/0244924 A1
Sheet 7 Multivariate testing results prediction model
Email designer interface empty state, drawing sheet 8 from US 2026/0244924 A1
Sheet 8 Email designer interface empty state
Show the other 22 sheets
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 9
Sheet 9
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 10
Sheet 10
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 11
Sheet 11
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 12
Sheet 12
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 13
Sheet 13
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 14
Sheet 14
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 15
Sheet 15
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 16
Sheet 16
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 17
Sheet 17
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 18
Sheet 18
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 19
Sheet 19
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 20
Sheet 20
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 21
Sheet 21
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 22
Sheet 22
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 23
Sheet 23
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 24
Sheet 24
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 25
Sheet 25
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 26
Sheet 26
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 27
Sheet 27
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 28
Sheet 28
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 29
Sheet 29
Adobe Patents an AI System That Picks Your Email Templates and Test Variants for You patent drawing, sheet 30
Sheet 30
Patent filing page

More in AI recommendation

All 86 AI recommendation patents →